Paper short abstract:

The presentation explores how new ways of “seeing” are emerging at the interface of root science and computer vision, where artificial neural networks promise open software and faster image segmentation.

Paper long abstract:

An increasing interest in soil matters stems from growing anxiety related to food production, destruction of the soils, and the threat imposed by climate change (Puig de la Bellacasa 2019; Counihan et al. 2020). An emerging line of literature in the social sciences is dedicated to the development of new conceptualizations of soil (Krzywoszynska and Marchesi, 2020), as well as fostering alternative approaches to human-soil relationships. These attempts are often positioned as overcoming prevailing technoscientific visions, which are seen as a continuation of extractive and abusive relationships between humans and soil. Yet in soil ecology and related disciplines, the shift in understanding soil as a process opens up new combinations and relations, challenging not only existing ways of knowledge production but turning soils into a ground for testing and contesting relations yet to be established. Drawing on ethnographic material from an EU Marie-Curie international training network experiment based in Iceland, my case study examines how novel digital representation and analysis of plant roots are not only creating potentiality for emerging infrastructures but also clash with existing epistemes of "seeing" and knowing soils stabilized within the community of root ecologists, reconfiguring expectations towards big data approaches and machine analysis of images.